U.S. patent application number 16/193444 was filed with the patent office on 2020-05-21 for method and system for health assessment of a track circuit and/or of a track section.
The applicant listed for this patent is ALSTOM Transport Technologies. Invention is credited to Jeffrey Fries, Nenad Mijatovic, Song Qin.
Application Number | 20200156674 16/193444 |
Document ID | / |
Family ID | 68583112 |
Filed Date | 2020-05-21 |
![](/patent/app/20200156674/US20200156674A1-20200521-D00000.png)
![](/patent/app/20200156674/US20200156674A1-20200521-D00001.png)
![](/patent/app/20200156674/US20200156674A1-20200521-D00002.png)
![](/patent/app/20200156674/US20200156674A1-20200521-D00003.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00001.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00002.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00003.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00004.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00005.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00006.png)
![](/patent/app/20200156674/US20200156674A1-20200521-M00007.png)
United States Patent
Application |
20200156674 |
Kind Code |
A1 |
Qin; Song ; et al. |
May 21, 2020 |
METHOD AND SYSTEM FOR HEALTH ASSESSMENT OF A TRACK CIRCUIT AND/OR
OF A TRACK SECTION
Abstract
A method and system for health assessment of a track circuit
and/or of a track section, the track circuit being configured for
detecting the presence of a vehicle on the track section between a
transmitting end and a receiving end of the track circuit. The
method, implemented by an electronic device, comprises obtaining,
from a sensor device placed near the receiving end, samples of an
electrical parameter of an electric signal transmitted between the
transmitting end and the receiving end of the track section,
forming a temporal series of received samples, applying an
automatic clustering algorithm to separate the received samples in
a predetermined number of clusters, selecting one of the clusters
and determining, for the selected cluster, a first peak value of
the received samples classified within the selected cluster, and
calculating a track circuit health indicator depending on the first
peak value determined for the selected cluster.
Inventors: |
Qin; Song; (Melbourne,
FL) ; Mijatovic; Nenad; (Melbourne, FL) ;
Fries; Jeffrey; (Grain Valley, MO) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
ALSTOM Transport Technologies |
Saint-Ouen |
|
FR |
|
|
Family ID: |
68583112 |
Appl. No.: |
16/193444 |
Filed: |
November 16, 2018 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
B61K 9/08 20130101; B61L
1/185 20130101; G06K 9/6223 20130101; G01R 31/50 20200101; B61L
1/20 20130101; B61L 1/188 20130101; B61L 1/187 20130101; B61L 1/186
20130101; B61L 1/18 20130101 |
International
Class: |
B61K 9/08 20060101
B61K009/08; G01R 31/02 20060101 G01R031/02 |
Claims
1. Method for health assessment of a track circuit and/or of a
track section, the track circuit being configured for detecting the
presence of a vehicle on the track section between a transmitting
end and a receiving end of the track circuit, the transmitting end
and the receiving end being associated to respective ends of the
track section, the method being implemented by an electronic
device, and comprising: obtaining, from a sensor device placed near
the receiving end, samples of an electrical parameter of an
electric signal transmitted between the transmitting end and the
receiving end of the track circuit, forming a temporal series of
samples, applying an automatic clustering algorithm to separate the
received samples in a predetermined number of clusters, selecting
one of said clusters and determining, for the selected cluster, a
first peak value of the received samples classified within the
selected cluster, and computing a track circuit health indicator
depending on the first peak value determined for the selected
cluster.
2. The method of claim 1, further comprising: obtaining, from a
sensor device placed near the transmitting end, samples of the
electrical parameter of the electric signal transmitted between the
transmitting end and the receiving end of the track circuit,
forming a temporal series of transmitted samples, applying the
automatic clustering algorithm to separate the transmitted samples
in a predetermined number of clusters, selecting one of said
clusters and determining, for the selected cluster, a second peak
value of the transmitted samples classified within the selected
cluster, wherein the track circuit health indicator further depends
on the second peak value.
3. The method of claim 1, wherein the automatic clustering
algorithm applied is a k-means algorithm.
4. The method of claim 3, further comprising, before applying the
k-means algorithm, automatically computing a number K of clusters
to use for classifying the temporal series of samples.
5. The method of claim 3, wherein the automatic clustering
algorithm comprises: computing a P-dimensional feature vector
associated to each sample or sub-group of samples of the temporal
series of samples.
6. The method of claim 5, wherein the automatic clustering
algorithm further comprises: a) obtaining K centroids, each
centroid being associated to a cluster, b) computing a distance
between each feature vector and each centroid, and assigning the
feature vector to a cluster associated to the closest centroid
according to the distance computed; c) for each cluster, obtaining
an updated centroid computed as a mean value of all feature vectors
assigned to the cluster associated to the centroid, d) repeating
steps b) and c) until a stop criterion is met.
7. The method of claim 1, further comprising obtaining a first
voltage value of the electric signal at the receiving end and a
second voltage value of the electric signal at the transmitting
end.
8. The method of claim 7, wherein the track circuit health
indicator is a ballast resistance and/or a rail resistance.
9. The method of claim 2, wherein the electrical parameter is a
current, the received samples are received current samples, the
transmitted samples are transmitted current samples, the first peak
value is a first peak current value and the second peak value is a
second peak current value.
10. The method of claim 9, wherein the track circuit health
indicator is either: a ballast resistance, expressed in Ohms per
thousand feet and computed by the following formula: R ballast = TL
.times. ( V tx + V rx ) 2 .times. 1000 .times. ( I tx - I rx )
##EQU00006## where TL is the track section length, V.sub.rx is a
first voltage value of the electric signal at the receiving end,
V.sub.tx is a second voltage value of the electric signal at the
transmitting end, I.sub.rx is the first peak current value and
I.sub.tx, is the second peak current value; or a rail resistance,
expressed in Ohms per feet and computed by the following formula: R
rail = 2 .times. ( V tx - V rx ) TL .times. ( I tx + I rx )
##EQU00007## where TL is the track section length, V.sub.rx is a
first voltage value of the electric signal at the receiving end,
V.sub.tx is a second voltage value of the electric signal at the
transmitting end, I.sub.rx is the first peak current value and
I.sub.tx, is the second peak current value.
11. System for health assessment of a track circuit and/or of a
track section, the track circuit being configured for detecting the
presence of a vehicle on the track section between a transmitting
end and a receiving end of the track circuit, the transmitting end
and the receiving end being associated to respective ends of the
track section, the system comprising an electronic device
comprising at least one processor configured to: obtain, from a
sensor device placed near the receiving end, samples of an
electrical parameter of an electric signal transmitted between the
transmitting end and the receiving end of the track circuit,
forming a temporal series of samples, apply an automatic clustering
algorithm to separate the received samples in a predetermined
number of clusters, select one of said clusters and determine, for
the selected cluster, a first peak value of the received samples
classified within the selected cluster, and compute a track circuit
health indicator depending on the first peak value determined for
the selected cluster.
12. The system of claim 11, wherein the processor is further
configured to: obtain, from a sensor device placed near the
transmitting end, samples of the electrical parameter of the
electric signal transmitted between the transmitting end and the
receiving end of the track circuit, forming a temporal series of
transmitted samples, apply the automatic clustering algorithm to
separate the transmitted samples in a predetermined number of
clusters, select one of said clusters and determine, for the
selected cluster, a second peak value of the transmitted samples
classified within the selected cluster, wherein the track circuit
health indicator further depends on the second peak value.
13. The system of claim 12, wherein the electrical parameter is a
current, the received samples are received current samples, the
transmitted samples are transmitted current samples, the first peak
value is a first peak current value and the second peak value is a
second peak current value.
Description
FIELD OF THE INVENTION
[0001] The present invention concerns a method and a system for
health assessment of a track circuit and/or of a track section, the
track circuit being designed for detecting the presence of a
vehicle on the track section.
[0002] The invention belongs to the field of railway operating and
maintenance.
BACKGROUND OF THE INVENTION
[0003] It is known that detecting the presence of vehicles, in
particular trains, on railway tracks is critical for railway
signaling systems, and is therefore an important component of
railway infrastructures security.
[0004] Track circuits are designed for detecting the presence of
vehicles, in particular trains, on tracks of a track section. A
track circuit is an electrical circuit using tracks and relays, for
transmitting DC electric signals from a first end, also called
transmitting end of the track circuit to a second end, also called
receiving end of the track circuit. When a train is present on a
track section belonging to the track circuit, the wheels and the
axle of the train induce a short-circuit, therefore no electric
signal is received at the receiving end of the track circuit,
indicating the presence of a train. Current sensor devices such as
ammeters are installed on both ends of the track circuit to record
the current values of the transmitted signal at the transmitting
end and the received signal at the receiving end of the track
circuit. The current values form samples of an electric signal. By
extracting different information from the recorded sample values,
it is possible to estimate the degradation/health condition of the
railway track section or track circuit, and calculate the accuracy
of train detection.
[0005] It is known to transmit a modulated signal in the form of
pulses with different pulse features in order to carry different
information, which causes sample values of the pulses received at
the receiving end to be heterogeneous. Furthermore, due to
environment conditions and track degradation, the received pulses
are also noisy.
[0006] The received noisy and heterogeneous samples are not
applicable for further data analysis, in particular cannot provide
a reliable assessment of the track circuit health.
[0007] There is a need to improve the track circuit health
assessment.
SUMMARY OF THE INVENTION
[0008] This and other objects are achieved by a method for health
assessment of a track circuit and/or of a track section, the track
circuit being configured for detecting the presence of a vehicle on
the track section between a transmitting end and a receiving end of
the track circuit, the method being implemented by an electronic
device, and comprising: [0009] obtaining, from a sensor device
placed near the receiving end, samples of an electrical parameter
of an electric signal transmitted between the transmitting end and
the receiving end of the track section, forming a temporal series
of samples, [0010] applying an automatic clustering algorithm to
separate the received samples in a predetermined number of
clusters, [0011] selecting one of said clusters and determining,
for the selected cluster, a first peak value of the received
samples classified within the selected cluster, and [0012]
computing a track circuit health indicator depending on the first
peak value determined for the selected cluster.
[0013] In embodiments of the invention, the method for health
assessment of a track circuit comprises one or more of the
following features, considered alone or according to all
technically possible combinations.
[0014] The method further comprises: [0015] obtaining, from a
sensor device placed near the transmitting end, samples of the
electrical parameter of the electric signal transmitted between the
transmitting end and the receiving end of the track circuit,
forming a temporal series of transmitted samples, [0016] applying
the automatic clustering algorithm to separate the transmitted
samples in a predetermined number of clusters, [0017] selecting one
of said clusters and determining, for the selected cluster, a
second peak value of the transmitted samples classified within the
selected cluster, wherein the track circuit health indicator
further depends on the second peak value.
[0018] The automatic clustering algorithm applied is a k-means
algorithm. The method further comprises, before applying the
k-means algorithm, automatically computing a number K of clusters
to use for classifying the temporal series of samples.
[0019] The automatic clustering algorithm comprises computing a
P-dimensional feature vector associated to each sample of the
temporal series of samples.
[0020] The automatic clustering algorithm further comprises: [0021]
a) obtaining K centroids, each centroid being associated to a
cluster, [0022] b) computing a distance between each feature vector
and each centroid, and assigning the feature vector to a cluster
associated to the closest centroid according to the distance
computed; [0023] c) for each cluster, obtaining an updated centroid
computed as a mean value of all feature vectors assigned to the
cluster associated to the centroid, [0024] d) repeating steps b)
and c) until a stop criterion is met.
[0025] The method further comprises obtaining a first voltage value
of the electric signal at the receiving end and a second voltage
value of the electric signal at the transmitting end.
[0026] The track circuit health indicator is a ballast resistance
and/or a rail resistance.
[0027] The electrical parameter is a current, the received samples
are received current samples, the transmitted samples are
transmitted current samples, the first peak value is a first peak
current value and the second peak value is a second peak current
value.
[0028] The track circuit health indicator is a ballast resistance,
expressed in Ohms per thousand feet and computed by the following
formula:
R ballast = TL .times. ( V tx + V rx ) 2 .times. 1000 .times. ( I
tx - I rx ) ##EQU00001##
[0029] where TL is the track section length, V.sub.rx is a first
voltage value of the electric signal at the receiving end, T.sub.tx
is a second voltage value of the electric signal at the
transmitting end, I.sub.rx is the first peak current value and
I.sub.tx, is the second peak current value.
[0030] The track circuit health indicator is a rail resistance,
expressed in Ohms per feet and computed by the following
formula:
R rail = 2 .times. ( V tx - V rx ) TL .times. ( I tx + I rx )
##EQU00002##
[0031] where TL is the track section length, V.sub.rx is a first
voltage value of the electric signal at the receiving end, V.sub.tx
is a second voltage value of the electric signal at the
transmitting end, I.sub.rx is the first peak current value and
I.sub.tx, is the second peak current value.
[0032] The invention relates also to a system for health assessment
of a track circuit and/or of a track section, the track circuit
being configured for detecting the presence of a vehicle on the
track section between a transmitting end and a receiving end of the
track circuit, comprising an electronic device comprising at least
one processor configured to:--obtain, from a sensor device placed
near the receiving end, samples of an electrical parameter of an
electric signal transmitted between the transmitting end and the
receiving end of the track section, forming a temporal series of
samples, [0033] apply an automatic clustering algorithm to separate
the received samples in a predetermined number of clusters, [0034]
select one of said clusters and determine, for the selected
cluster, a first peak value of the received samples classified
within the selected cluster, and [0035] compute a track circuit
health indicator depending on the first peak value determined for
the selected cluster.
BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Further characteristics and advantages of the present
invention will become apparent from the following description,
provided merely by way of non-limiting example, with reference to
the enclosed drawings, in which:
[0037] FIG. 1 shows a track circuit with a corresponding track
section and a system for health assessment according to an
embodiment of the present invention;
[0038] FIG. 2 shows an example of a series of received current
samples classified into two clusters;
[0039] FIG. 3 shows steps of a method for health assessment of a
track circuit according to an embodiment of the present
invention.
DETAILED DESCRIPTION OF EMBODIMENTS
[0040] FIG. 1 illustrates schematically an example of a track
section 2 belonging to a track circuit 4, and a health assessment
system 6 according to an embodiment of the invention.
[0041] The track section 2 is a portion of a railway track, for
example formed by two parallel rails 2A and 2B as shown, suitable
for forming a road for railway vehicles, e.g. trains. The rails 2A,
2B are fixed using railway sleepers 8 placed on ballast 10.
[0042] Classically, the rails 2A, 2B are made of an electrically
conducting metal, for example steel.
[0043] The track section 2 is comprised between a first insulating
joint 12 and a second insulating joint 14. In the example shown,
each insulating joint 12, 14 encompasses the two parallel rails 2A
and 2B forming the track.
[0044] The first insulating joint 12 defines a first end, also
called transmitting end T.sub.E of the track circuit 4 and the
second insulating joint 14 defines a second end, also called
receiving end R.sub.E of the track circuit 4.
[0045] The track circuit 4 further comprises an electric signal
generator and modulator 20 schematically represented as a single
unit, configured to generate, from a direct current (abbreviated as
DC) electric signal power source, modulated electric signals, coded
to carry information. The modulated electric signals are obtained
by turning on and off a power generator at a given rate called
modulation frequency, the purpose of modulation being to carry
information. A modulated electric signal is formed by a series of
pulses.
[0046] The pulses are characterized by several pulse features:
pulse center frequency, pulse bandwidth, pulse duration or width,
and time between pulses.
[0047] In an embodiment, two different pulse widths are applied and
therefore two types of pulses are generated, and transmitted.
[0048] The electric signal generator and modulator 20 is connected
by two conducting branches to each conducting rail 2A, 2B.
[0049] The track circuit 4 may include switches and relays which
are conventional and are not represented in the FIG. 1.
[0050] When there is no train on the track section 2, an electric
signal circulates in the circuit created by the connection of the
generator and modulator 20 to the conducting rails of the track
section, between the first insulating joint 12 and the second
insulating joint 14. A modulated electric signal transmitted from
the transmitting end T.sub.E is received at the receiving end
R.sub.E.
[0051] When a train is present of the track section 2, the wheels
and axle of the train achieve a short-circuit and no electric
signal is received at the receiving end R.sub.E of the track
circuit 4.
[0052] In order to measure the current amplitude, two current
sensor devices 22 and 24, for example ammeters, are placed
respectively near the transmitting end T.sub.E of the track circuit
and near the receiving end R.sub.E of the track circuit, i.e. at a
given proximity to the respective insulating joints 12 and 14. For
example, each current sensor device is placed at a distance
comprised between 0 meter to 10 meters from the insulating
joint.
[0053] Each current sensor device is configured to measure temporal
series of current samples, corresponding respectively to the
current values of the transmitted or received electrical
pulses.
[0054] In an embodiment, the modulated electric signal transmitted
between the transmitting end and the receiving end of the track
circuit is composed of several types of pulses generated with
different pulse features.
[0055] In some embodiments, the track circuit 4 includes several
types of sensors such as current, voltage and power sensors. Each
sensor device is configured to measure temporal series of current,
voltage or power samples.
[0056] Preferably, the sample values are captured during a period
of given duration, for example, 2 seconds, and are in addition
uniformly sampled (with minimal resolution of 1 ms) and ordered in
the form of temporal series of samples values.
[0057] Depending on the embodiments, the sampling and ordering
operations are performed either by the measuring devices or in Data
Acquisition Units (DAUs) 30, 32 of the health assessment system
which in this case include an analogue-to-digital converter and a
processor.
[0058] FIG. 2 shows a temporal series S of current samples obtained
between a first instant T.sub.0 and a second instant T.sub.F,
forming a period of 270 ms, the horizontal axis showing the time
(in milliseconds) and the vertical axis showing current values (in
A).
[0059] In the example, the series S is actually composed of current
samples from two types of pulses of different pulse widths.
[0060] It is proposed to apply a k-means clustering algorithm to
automatically classify the current samples into two clusters,
referenced Cluster 1 and Cluster 2 in this example and indicated
with a triangular reference for the samples belonging to Cluster 1
and a disk reference for the samples belonging to Cluster 2.
[0061] The clustering algorithm is applied by the health assessment
system 6, an embodiment of which is shown in FIG. 1.
[0062] The health assessment system 6 is an electronic system
comprising one or several Data Acquisition Units (DAU). In the
example of FIG. 1, the health assessment system 6 comprises two DAU
30, 32.
[0063] The DAU 30 is configured to collect current samples from the
current sensor device 22 placed at the transmitting end of the
track circuit. These current samples are called transmitted current
samples.
[0064] The DAU 32 is configured to collect current samples from the
current sensor device 24 placed at the receiving end of the track
circuit. These current samples are called received current
samples.
[0065] Each DAU 30, 32 comprises connectors through which the data
exchange is performed, such as USB ("Universal Serial Bus")
connectors, radiofrequency connectors such as WIFI, Bluetooth or
modems to connect to Internet.
[0066] The collected series of current samples are stored in a
database 34.
[0067] The health assessment system 6 further comprises a k-means
clustering unit 36, a peak detector unit 38 and a health indicator
computing unit 40.
[0068] In an embodiment, the k-means clustering unit 36, peak
detector unit 38 and health indicator computing unit 40 are
implemented as software, stored in a memory 42 of the health
assessment electronic system 6, and executed by the processor 44.
Alternatively, the units 36, 38 and 40 are stored on a non-volatile
information recording medium, such as an optical disk, a
magneto-optical disk, any type of non-volatile memory (e.g. EPROM,
EEPROM, FLASH, NVRAM), a magnetic card or and optical card.
[0069] In an alternative embodiment, each of the units 36, 38 and
40 is implemented by a FPGA (Field Programmable Gate Array), or a
dedicated integrated circuit such as an ASIC (Applications Specific
Integrated Circuit).
[0070] The health assessment electronic system 6 is configured to
transmit the computed health indicators to a rail maintenance
center 46. In an embodiment, the transmission is carried out via
radiofrequency connectors, using a wireless communication protocol.
In a variant, the health assessment electronic system 6 is situated
within a rail maintenance center and the transmission is carried
out by a wired connection.
[0071] A health indicator of the track circuit is a parameter
having a value representative of a level of degradation of the
track circuit 4 or of a part of the track circuit 4, for example
the track section 2.
[0072] The computed health indicators are used to apply predictive
maintenance, for example by recording successive values of each
health indicator computed is it possible to accurately predict a
track-circuit failure or a track section degradation before an
actual failure, and to apply corrective measures to prevent
failures.
[0073] FIG. 3 shows steps of a method for health assessment of a
track circuit according to an embodiment.
[0074] In the embodiment of FIG. 3, analogous steps, performed on
temporal series of current samples from the transmitting end and
from the receiving end, are shown.
[0075] In a first step 100, a temporal series of current samples
are obtained from a current sensor device placed near the receiving
end.
[0076] An analogous step 110 is carried out to obtain a temporal
series of current samples from a current sensor device placed near
the transmitting end.
[0077] For example, in an embodiment, the samples are obtained from
payload packets formatted according to a communication protocol,
for example UDP ("User Datagram Protocol") packets. Each UDP packet
consists of 1 to 2 pulses, where each pulse consists of a number of
temporal samples, for example, 43.
[0078] Next, the temporal series of samples is processed in feature
extraction step 120 to obtain estimated pulse features.
[0079] The processing comprises comparing each current sample value
to a current threshold value. For example, for current samples
comprised between 0 and 3.5 A, the threshold value is set to 1
A.
[0080] The current sample values above the threshold are used to
compute estimated peak amplitude (maximum value of samples above
the threshold, for example, 2.3 A) and pulse width (number of
samples above the threshold, for example, 20). One peak amplitude
and one pulse width are computed for each pulse. To carry out a
clustering algorithm, we need a minimum of 1000 to 2000 pulses.
[0081] In a variant, before estimating the pulse features, train
moves data is filtered out, based on a track occupancy flag
provided by the track circuit software. For example, a flag with
value 1 indicates that the track is occupied by train and a flag
with 0 indicates that the track is not occupied.
[0082] A feature extraction step 130, analogous to step 120, is
carried to obtain estimated features from a temporal series of
current samples from the current measuring sensor placed near the
transmitting end.
[0083] Step 120 (respectively 130) is followed by an automatic
clustering step 140 (respectively 150), applied by the k-means
clustering unit 36, to separate the received current samples into a
predetermined number of clusters, each cluster corresponding to
current samples of pulse of a same type.
[0084] In the preferred embodiment k-means clustering is applied,
with a predetermined number of clusters K.
[0085] The k-means clustering algorithm is applied on P-dimensional
vectors of features, each vector of feature being associated with a
current sample or a sub-group of current samples of the series of
current samples, with P being a positive integer. For example, K=2
for 2 clusters.
[0086] For example, in a first embodiment, only the estimated pulse
width is used (P=1). In a second embodiment, the peak amplitude and
the pulse width are used (P=2).
[0087] The k-means algorithm comprises the following steps: [0088]
a) Obtaining K centroids, which are for example randomly chosen at
initializing. Each centroid is a P-dimensional vector of feature
which is the center of an associated cluster; [0089] b) Each
feature vector is assigned to the cluster associated to the
centroid that is closest according to a predetermined distance. For
example, Euclidean distance is used. Other distance metrics include
Manhattan distance which is computed as absolute difference between
coordinates of pair of objects. [0090] c) The mean vector of each
cluster is updated after all samples are classified, so as to
obtain K updated centroids; [0091] d) The steps b) and c) are
repeated until the centroids positions converge in the
P-dimensional space, and more generally until a stop criterion is
met. For example, the stop criterion may be a predetermined number
of iterations. In an alternative embodiment, the stop criterion may
be ensuring that the distance in the P-dimensional space of two
successive updated centroids associated to the same cluster is less
than a given threshold.
[0092] The result of the automatic clustering is to obtain clusters
of homogeneous samples.
[0093] Next, one of the clusters is selected, and a peak current
value is computed (step 160, 170) by peak detector unit 38. The
peak current value is the maximum current sample value of all
samples belonging to the selected cluster.
[0094] For example, in an embodiment, the cluster with the smallest
centroid value is chosen for the processing at the transmitting end
and at the receiving end.
[0095] Referring to FIG. 2, for example Cluster 2 is selected, and
the peak current value is computed (for example, the peak current
value is around 2.75 A in the example of FIG. 2).
[0096] One or several health indicators are computed by the health
indicator computing unit 40 in next health indicator computing step
180.
[0097] Ballast resistance is one of the indicators of the health
condition of the track circuit computed in step 180. A low ballast
resistance corresponds to a bad health condition, and a high
ballast resistance corresponds to good health condition.
[0098] The definition of "low resistance" and "high resistance" is
obtained for example from experimental data, collected from healthy
track sections/track circuits, for example on a new material. The
ballast resistance, expressed in Ohms/1000 feet, is given by the
formula:
R ballast = TL .times. ( V tx + V rx ) 2 .times. 1000 .times. ( I
tx - I rx ) ( 1 ) ##EQU00003##
[0099] where TL is the track section length, V.sub.rx is a first
voltage value of the modulated electric signal at the receiving
end, V.sub.tx is a second voltage value of the modulated electric
signal at the transmitting end, I.sub.rx is the peak current value
at the receiving end as computed in step 160, and I.sub.tx, is the
peak current value at the transmitting end as computed in step
170.
[0100] V.sub.tx is given by the formula:
V.sub.tx=V.sub.s-(I.sub.tx.times.R.sub.tx) (V) (2)
where V.sub.s is the voltage of electric signal generator 20 and
R.sub.tx is the resistance of the wires connecting the track
circuit and the rails at the transmitting end.
[0101] V.sub.rx is given by the formula:
V.sub.rx=I.sub.rx.times.R.sub.rx(V) (3)
where R.sub.rx is the resistance of the wires connecting the track
circuit and the rails at the receiving end.
[0102] For example, the ballast resistance of a track circuit with
the following parameters is 7.6 (Ohms/1000 ft)-- [0103] TL=9210
(feet), [0104] V.sub.s=2.0 (V), [0105] R.sub.tx=R.sub.rx=0.53
(Ohms) [0106] I.sub.tx=2.46 (A) [0107] I.sub.rx=1.54 (A)
[0108] Rail resistance is another one of the indicators of the
health condition of the track circuit computed in step 180. An
abnormal rail resistance might correspond to a broken rail
condition. The "normal" rail resistance values are obtained for
example from experimental data, collected from healthy/new track
sections.
[0109] The rail resistance, expressed in Ohms/feet, is given by the
formula:
R rail = 2 .times. ( V tx - V rx ) TL .times. ( I tx + I rx ) ( 4 )
##EQU00004##
[0110] For example, the rail resistance of a track circuit with the
following parameters is -6.5E-06 (Ohms/ft)-- [0111] TL=9210 (feet),
[0112] V.sub.s=2.0 (V), [0113] R.sub.tx=R.sub.rx=0.53 (Ohms) [0114]
I.sub.tx=2.46 (A) [0115] I.sub.rx=1.54 (A)
[0116] In an embodiment, either only the ballast resistance or only
the rail resistance is computed. In a variant, both the ballast
resistance and the rail resistance are computed.
[0117] In another variant, further track circuit/track section
health indicators are computed.
[0118] The health indicator(s) are then processed in step 190, for
example transmitted to a rail maintenance center and compared to
recorded threshold values in order to detect problems and estimates
risks of failure, and to apply predictive maintenance.
[0119] In an alternative embodiment, the values of voltage and
current at the transmitting end of the track circuit are previously
stored, and only the values of the voltage and of the peak current
amplitude are measured. In this alternative, only the steps 100,
120, 140 and 160 are applied, whereas the peak current amplitude at
the transmitting end is obtained from a memory of the health
assessment system.
[0120] In another alternative embodiment, the number K of clusters
applied in the K-means clustering algorithm is not known in
advance, but is estimated automatically. For example, several
values for K are tested with P-dimensional vectors of features, for
example k varying from 1 to 9, and the best K value is chosen based
on the within-cluster sum of squares according to the formula:
S ( k ) = i = 1 k x .di-elect cons. C i x - .mu. i 2 ( 5 )
##EQU00005##
[0121] Where C.sub.i is the i.sup.th cluster among k clusters, and
is the centroid of cluster C.sub.i, and x is a vector of features
belonging to C.sub.i.
[0122] The computed values S(k) can be plotted, and the chosen
value for k is the value K for which the computed plot forms an
elbow.
[0123] In particular, computing the number of clusters can be
applied each time the type of pulses transmitted is changed.
[0124] The invention has been described in an embodiment using
samples of current pulses of the electric signal transmitted
between the transmitting end and the receiving end of the track
section. In alternative embodiments, the method for health
assessment of a track circuit and/or a track section applies with
samples of other electrical parameters of the electric signal
transmitted between the transmitting end and the receiving end of
the track section.
[0125] Advantageously, the proposed method does not require
building any reference on the features characterizing the pulses
and can adapt automatically to changes in terms of pulse
features.
[0126] Advantageously, the proposed method achieves automatic
clustering of samples belonging to pulses, and therefore the data
sets used to compute health indicators are less noisy, and the
health indicators computed are more reliable.
* * * * *